PRECIPITATING FACTORS AND THEIR ASSOCIATION WITH SEVERITY OF DIABETIC KETOACIDOSIS IN TYPE 1 DIABETES MELLITUS: A PROSPECTIVE OBSERVATIONAL STUDY

Background: Diabetic ketoacidosis (DKA) is the most serious acute metabolic complication of type 1 diabetes mellitus (T1DM). Most episodes follow an identifiable precipitant, but the relative contribution of individual precipitants and their bearing on severity differ appreciably between populations. This study examined the precipitants of DKA and their association with severity at a tertiary centre in Central India. Methods: A prospective observational study was conducted over 18 months in the Department of General Medicine, Gandhi Medical College and Hamidia Hospital, Bhopal. Sixty consecutive patients older than 14 years with T1DM presenting in DKA were enrolled. Precipitants were established from clinical assessment, laboratory investigation, microbiology and imaging; severity was graded from pH and bicarbonate. Proportions are reported with 95% Wilson confidence intervals. Because severity is an ordered variable, its association with each precipitant was assessed by proportional-odds ordinal logistic regression, with an exact conditional test for linear trend as the test of significance. Results: Mean age was 26.8 ± 8.2 years and 34 patients (56.7%) were male. Infection was the commonest precipitant, in 34 patients (56.7%, 95% CI 44.1-68.4), followed by insulin omission or poor adherence in 16 (26.7%, 95% CI 17.1-39.0) and new-onset T1DM in 10 (16.7%, 95% CI 9.3-28.0). Pneumonia (12/34, 35.3%) and urinary tract infection (10/34, 29.4%) were the leading infectious foci; tuberculosis accounted for a further 6 (17.6%). Moderate DKA was commonest (22, 36.7%), then mild (20, 33.3%) and severe (18, 30.0%). Severe DKA occurred in 41.2% of infection-precipitated episodes against 12.5% after insulin omission. Infection was associated with higher severity (proportional-odds ratio 3.18, 95% CI 1.19-8.48; exact trend p = 0.023). Insulin omission showed an inverse association of borderline significance (OR 0.33, 95% CI 0.11-0.99; exact p = 0.067), and new-onset T1DM showed none (OR 0.64, 95% CI 0.18-2.21; p = 0.524). Conclusion: Infection was the leading precipitant of DKA in this cohort and was associated with roughly threefold higher odds of more severe disease. Infection and insulin omission together accounted for five-sixths of episodes and both are largely preventable, so prompt treatment of intercurrent infection, sustained insulin adherence, structured sick-day guidance and earlier recognition of T1DM are the practical priorities.